Project plan Gantt chart generation method and device, computer equipment and medium

Abstract

The invention relates to the technical field of natural language processing, and particularly discloses a project plan Gantt chart generation method and device, computer equipment and a medium. According to the method, the initial management process can be automatically generated according to the to-be-analyzed file and the business knowledge graph, the process is adjusted according to the execution result of each process node in the execution process, and the project plan Gantt chart is generated, so that automatic generation and dynamic optimization of the project plan Gantt chart are realized, manual participation is not needed, and the efficiency is improved. And the generation efficiency of the project plan Gantt chart is improved. The method is applied to management process generation business of projects such as financial product development and medical tool research and development, the generation efficiency of the project management process can be improved, the management process can be dynamically optimized according to the execution result of each process node, and the management efficiency of the projects such as product development and tool research and development is improved.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, and in particular to a method, apparatus, computer device, and storage medium for generating a project plan Gantt chart. Background Technology

[0002] With the rapid development of science and technology, development projects in the financial and medical fields are increasing daily, such as insurance product development, drug research and development, and medical tool development. These projects face the dual challenges of agile development and compliance risk control. Traditional project management and compliance management processes require manual adaptation to project management processes and compliance requirements, which is inefficient and prone to errors. This is especially true in scenarios such as insurance product development and fund management, where differences in regulatory policies across different regions lead to complex process configurations, resulting in low efficiency in generating project plan Gantt charts. This fails to meet the dual requirements of agility and compliance for product development and tool development. Therefore, improving the efficiency of project plan Gantt chart generation has become an urgent problem to be solved. Summary of the Invention

[0003] This application provides a method, apparatus, computer equipment, and storage medium for generating project plan Gantt charts, so as to improve the efficiency of project plan Gantt chart generation.

[0004] Firstly, this application provides a method for generating a project plan Gantt chart, the method comprising: Obtain the file to be parsed and the business knowledge graph corresponding to the target management project, and process the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project; Based on the initial management process, a smart contract is generated and executed to obtain the execution results of each process node in the initial management process. The execution results are then risk-scored based on a risk scoring model to obtain the compliance risk score of each process node in the initial management process. Obtain the management parameters corresponding to the target management project, analyze the management parameters and the compliance risk score based on the reinforcement learning model, adjust the process according to the analysis results, and generate a project plan Gantt chart.

[0005] Secondly, this application also provides an apparatus for generating a project plan Gantt chart, the apparatus comprising: The initial management process generation module is used to obtain the file to be parsed and the business knowledge graph corresponding to the target management project, and process the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project. The compliance risk score acquisition module is used to generate and execute smart contracts based on the initial management process, obtain the execution results of each process node in the initial management process, and perform risk scoring on the execution results based on the risk scoring model to obtain the compliance risk score of each process node in the initial management process. The project plan Gantt chart generation module is used to obtain the management parameters corresponding to the target management project, analyze the management parameters and the compliance risk score based on the reinforcement learning model, and adjust the process according to the analysis results to generate the project plan Gantt chart.

[0006] Thirdly, this application also provides a computer device, the computer device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the method for generating a project plan Gantt chart as described above.

[0007] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the method for generating a project plan Gantt chart as described above.

[0008] This application discloses a method, apparatus, computer device, and storage medium for generating a project plan Gantt chart. The method involves acquiring a file to be parsed and a business knowledge graph corresponding to a target management project, processing the file and the knowledge graph to obtain an initial management process for the target project, generating and executing a smart contract based on the initial management process, obtaining the execution results of each process node in the initial management process, and performing a risk score on the execution results based on a risk scoring model to obtain a compliance risk score for each process node in the initial management process. The method also involves acquiring management parameters corresponding to the target management project, analyzing the management parameters and the compliance risk score based on a reinforcement learning model, and adjusting the process based on the analysis results to generate the project plan Gantt chart. This application can automatically generate an initial management process based on the file to be parsed and the business knowledge graph, and adjust the process based on the execution results of each process node during execution to generate the project plan Gantt chart. This achieves automated generation and dynamic optimization of the project plan Gantt chart without manual intervention, thus improving the generation efficiency of the project plan Gantt chart. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic flowchart of a method for generating a project plan Gantt chart provided in the first embodiment of this application; Figure 2 This is a schematic flowchart of a method for generating a project plan Gantt chart provided in the second embodiment of this application; Figure 3 This is a schematic flowchart of a method for generating a project plan Gantt chart provided in the third embodiment of this application; Figure 4 A schematic block diagram of a project plan Gantt chart generation apparatus provided for embodiments of this application; Figure 5 A schematic block diagram of the structure of a computer device provided for an embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0013] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0015] This application provides a method, apparatus, computer device, and storage medium for generating project plan Gantt charts. The method for generating project plan Gantt charts can be applied to a server, automatically generating an initial management process based on the file to be parsed and a business knowledge graph. During execution, the process is adjusted based on the execution results of each process node to generate the project plan Gantt chart. This achieves automated generation and dynamic optimization of project plan Gantt charts without manual intervention, improving the generation efficiency. The server can be a standalone server or a server cluster.

[0016] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0017] Please see Figure 1 , Figure 1 This is a schematic flowchart illustrating a method for generating a project plan Gantt chart, as provided in an embodiment of this application. This method can be applied to a server to automatically generate an initial management process based on the file to be parsed and a business knowledge graph. During execution, the process is adjusted based on the execution results of each process node to generate the project plan Gantt chart. This achieves automated generation and dynamic optimization of the project plan Gantt chart without manual intervention, thus improving the efficiency of project plan Gantt chart generation.

[0018] like Figure 1 As shown, the method for generating the Gantt chart for this project specifically includes steps S101 to S103.

[0019] S101. Obtain the file to be parsed and the business knowledge graph corresponding to the target management project, and process the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project. In one embodiment, the document to be parsed can be a policy document such as regulatory provisions or methods of financial regulatory agencies or drug development regulations of medical regulatory agencies, or it can be a project document within a financial enterprise or medical institution, such as a project plan or requirements document, which contains relevant information on project management, such as project objectives, task lists, compliance clauses, product terms, timelines, and risk control rules.

[0020] A business knowledge graph is a structured knowledge base that contains various concepts, relationships, and rules in the project management domain. For example, a business knowledge graph can define dependencies between different tasks, rules for resource allocation, and so on.

[0021] In one embodiment, key information such as project name, compliance terms, and regulatory rules are extracted from the file to be parsed using natural language processing technology. Key entities are extracted from the business knowledge graph, and the logical relationships between the key entities are identified to obtain compliance constraints and process nodes to be processed. The compliance constraints are then mapped to the process nodes to be processed to obtain the initial management process.

[0022] Furthermore, the step of obtaining the business knowledge graph corresponding to the target management project includes: parsing the business files corresponding to the target management project based on the local large language model of each project manager to obtain at least one key entity matching the target management project; identifying the relationships between the key entities based on a dependency parsing model to obtain the logical relationships between the key entities; and aggregating the received key entities and the logical relationships between the key entities based on a federated learning algorithm to generate the business knowledge graph.

[0023] In one embodiment, business documents may include project plans, requirements documents, contracts, meeting minutes, etc., containing relevant information for project management.

[0024] In one embodiment, each project manager has a trained local large language model adapted to specific business domains or project needs. The local large language model is used to parse business documents and extract key information. Specifically, the business documents undergo text preprocessing, including text cleaning (removing irrelevant characters, punctuation, etc.) and word segmentation (dividing the text into words or phrases). The preprocessed text is then used to identify entities in the document, such as project names, task names (e.g., cross-border data transfer), responsible persons, timelines, resource names, and clauses. The extracted entities are then semantically analyzed and matched with the target management project to ensure that the extracted entities are relevant to the target management project. For example, if the target management project is "Project A," only entities related to "Project A" are extracted.

[0025] In one embodiment, dependency parsing is a natural language processing technique used to analyze dependency relationships between words in a sentence. Dependency relationships can reveal logical and semantic relationships between words. The dependency parsing model is obtained by training a pre-trained model using a collected training dataset.

[0026] By using a dependency parsing model, logical relationships between key entities are identified. These logical relationships include task dependencies, such as "Task 1" must be completed before "Task 2"; and time sequence relationships, such as "Task 1" should be completed before "October 1, 2025".

[0027] In one embodiment, each project manager sends the locally parsed key entities and logical relationships to a central server. The central server aggregates this data using a federated learning algorithm. Specifically, first, an entity alignment tool from natural language processing is used to align entities from different project managers, ensuring that identical entities extracted by different managers are correctly aligned. For example, different managers might use different names to refer to the same task, requiring entity alignment to unify them. Then, the logical relationships extracted by different managers are merged to generate a complete logical relationship network, constructing a business knowledge graph. A knowledge graph is a structured knowledge base that represents key entities and the relationships between them in graph form.

[0028] In one embodiment, a knowledge graph typically contains nodes (entities) and edges (relationships). Each node represents a key entity, and each edge represents a logical relationship between entities.

[0029] In the above embodiments, federated learning is used to parse business documents using the local large language models of each project management party, extract key entities and logical relationships, and then aggregate them through a federated learning algorithm to generate a business knowledge graph. This not only makes full use of the local data and models of each management party, but also builds a comprehensive business knowledge graph while protecting data privacy, providing strong support for project management.

[0030] Further, the step of processing the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project includes: parsing the file to be parsed based on a preset large language model to obtain a first compliance constraint and a first process node; extracting information from the business knowledge graph to obtain a second compliance constraint and a second process node; integrating the first and second compliance constraints to generate compliance constraints; analyzing and integrating the first and second process nodes to generate a process node to be processed; and mapping the compliance constraints to the process node to be processed to obtain the initial management process.

[0031] In one embodiment, a large language model is used to parse the document to be parsed and extract key information. Specifically, the document undergoes text preprocessing, including text cleaning (removing irrelevant characters, punctuation marks, etc.) and word segmentation (dividing the text into words or phrases). Clause identification is then performed on the preprocessed text, using the compliance constraints extracted from the document as the primary compliance constraints, including legal requirements, industry standards, and regulatory clauses. For example, the document may stipulate that certain tasks must be completed within a specific timeframe, or that certain tasks must be performed by personnel with specific qualifications.

[0032] The process nodes extracted from the document serve as the second process nodes, defining key steps in the project management process, such as the start and end of tasks and the setting of milestones.

[0033] In one embodiment, compliance constraints and process nodes related to the target management project are extracted from the business knowledge graph. Specifically, graph traversal algorithms (such as breadth-first search or depth-first search) can be used to extract information from the graph.

[0034] Compliance constraints extracted from the knowledge graph, as secondary compliance constraints, can include industry best practices, historical project experience, etc.

[0035] Process nodes extracted from the knowledge graph serve as second process nodes, defining standard steps and best practices in the project management process.

[0036] In one embodiment, a first compliance constraint extracted from a document and a second compliance constraint extracted from a knowledge graph are merged to obtain the final compliance constraints. This includes removing duplicate conditions and resolving potential conflicts. For example, if the conditions in the document and the conditions in the knowledge graph are inconsistent, the priority of the user-defined document and knowledge graph determines whether to retain the compliance constraints corresponding to the document or the compliance constraints in the knowledge graph.

[0037] In one embodiment, a first process node extracted from a file and a second process node extracted from a knowledge graph are merged to obtain a process node to be processed. This includes removing duplicate nodes and optimizing the order and dependencies of the process nodes. For example, if the task order defined in the file is inconsistent with the best practices in the knowledge graph, the priority of the user-defined file and knowledge graph is used to determine whether to retain the process node corresponding to the file or the process node in the knowledge graph.

[0038] In one embodiment, the obtained compliance constraints are mapped to pending process nodes to obtain the initial management process. The initial management process is a complete project management process that defines all steps from the start to the end of the project and ensures that each step complies with compliance constraints.

[0039] In one embodiment, the initial management process can be visualized in the form of Gantt charts, flowcharts, etc., to facilitate management and monitoring by the project team.

[0040] Further, the step of mapping the compliance constraints to the process nodes to obtain the initial management process includes: classifying the compliance constraints according to data modalities to obtain text data, structured data, and knowledge graph data; mapping the text data, structured data, and knowledge graph data to a shared vector space to obtain a unified standard data vector; and performing sparse attention processing on the data vector and the process nodes to be processed based on a cross-attention layer in a preset self-attention model to match and map the compliance constraints to the process nodes to be processed, thereby obtaining the initial management process.

[0041] In one embodiment, a data modality refers to different forms and types of data. In this embodiment, compliance constraints are divided into three main data modalities: text data, which exists in the form of natural language, such as contract terms, legal provisions, project requirement documents, etc.; structured data, which exists in the form of tables, such as records in a database, Excel spreadsheets, etc., which usually contain explicit fields and values; and knowledge graph data, which exists in the form of a graph structure, containing nodes (entities) and edges (relationships), such as task dependencies and resource allocation rules in a business knowledge graph.

[0042] Specifically, natural language processing techniques are used to extract key information from text, such as entities and relationships. Data parsing techniques are used to extract fields and values ​​from tables, ensuring data integrity and accuracy. Graph traversal algorithms are used to extract nodes and edges from the knowledge graph, ensuring the integrity and consistency of the graph structure.

[0043] In one embodiment, the shared vector space is a unified vector space that allows data from different modalities to be represented in the same form.

[0044] In a specific embodiment, a large language model is used to convert text data into vector representations. Feature engineering methods are used to convert structured data into vector representations; for example, fields and values ​​in a table are converted into feature vectors. Graph embedding techniques are used to convert nodes and edges in a knowledge graph into vector representations. Through these steps, data from different modalities are converted into a unified vector form, enabling data from different modalities to be processed in the same vector space.

[0045] In one embodiment, the self-attention model is obtained by training a pre-trained model using historical data. The cross-attention layer is used to handle the attention relationship between two different sequences. In this embodiment, it is used to handle the relationship between data vectors and nodes in the process to be processed.

[0046] Sparse attention is a variant of attention mechanisms that makes models more efficient by limiting the scope of attention. Sparse attention can reduce computational cost while improving model interpretability.

[0047] The data vector and the nodes to be processed are used as input to the self-attention model. Attention weights between the data vector and the nodes are calculated using a cross-attention layer. A sparse attention mechanism is then applied to perform sparse multimodal dimensionality reduction and compress high-dimensional features, removing redundant information such as "policy issuing unit" and "node leader," reducing interference from redundant information, minimizing unnecessary attention calculations, and improving efficiency.

[0048] In a specific embodiment, the data vectors corresponding to text data, structured data, and knowledge graph data are combined to generate a feature vector matrix, which is then sparsely decomposed and mapped to a low-dimensional space to achieve dimensionality reduction. The semantic similarity between the dimensionality-reduced process nodes and compliance constraints is calculated. Using a softmax function, the semantic similarity is converted into attention weights; higher weights indicate a greater relevance between the compliance clause and the process node. The attention weights and compliance clauses are weighted and summed to obtain the aligned compliance constraint vector, which serves as an appendix attribute of the process node. This mapping of compliance conditions to process nodes ensures that each process node has compliance requirements to meet.

[0049] In the above embodiments, by combining a pre-set large language model and a business knowledge graph, information is extracted from documents and the knowledge graph, integrated, and mapped to generate an initial management process. This fully utilizes the specific information in the documents and the structured knowledge in the knowledge graph to automatically generate the initial management process. Furthermore, the sparse attention mechanism improves processing efficiency and the overall efficiency of management process generation. S102. Generate and execute a smart contract based on the initial management process to obtain the execution results of each process node in the initial management process, and score the execution results based on a risk scoring model to obtain the compliance risk score of each process node in the initial management process. In one embodiment, the rules and process logic of project management are translated into smart contract code based on the initial management process. A smart contract is a contract with automatically executed terms stored on a blockchain. The contract content may include the start and end conditions of tasks, resource allocation rules, dependencies between tasks, etc.

[0050] In one embodiment, when certain conditions are triggered during the project management process, the smart contract automatically executes the corresponding operation. For example, when one task is completed, the start of the next task is automatically triggered.

[0051] The execution results of smart contracts are recorded on the blockchain, including the task completion time and compliance verification results.

[0052] In one embodiment, the risk scoring model is a data analytics and machine learning-based model used to assess risks in the project management process, trained using historical data. Based on the execution results of smart contracts, the risk scoring model assesses the risk of each process node, obtaining a compliance risk score for each node. The compliance risk score reflects the compliance and risk level of each process node.

[0053] S103. Obtain the management parameters corresponding to the target management project, analyze the management parameters and the compliance risk score based on the reinforcement learning model, adjust the process according to the analysis results, and generate a project plan Gantt chart.

[0054] In one embodiment, management parameters include project type and team capability data of the project management team.

[0055] Reinforcement learning models are machine learning models that learn optimal behavioral strategies through interaction with their environment. In project management, the environment can be the project management process, and the behavioral strategy can be the decision to adjust the process.

[0056] The reinforcement learning model, based on management parameters and compliance risk scores, outputs how to adjust project management processes to reduce risk and improve efficiency. Examples include adjusting task order and inserting compliance review milestones.

[0057] In one embodiment, a Gantt chart is a commonly used project management tool for displaying the timeline and progress of project tasks. A project plan Gantt chart is generated based on the adjusted management process. The Gantt chart clearly shows the start and end times of each task, dependencies between tasks, etc., facilitating management and monitoring by the project team.

[0058] The above embodiments provide a method, apparatus, computer equipment, and storage medium for generating project plan Gantt charts. The method involves acquiring a file to be parsed and a business knowledge graph corresponding to a target management project, processing the file to be parsed and the business knowledge graph to obtain an initial management process corresponding to the target management project, generating and executing a smart contract based on the initial management process to obtain the execution results of each process node in the initial management process, and performing a risk score on the execution results based on a risk scoring model to obtain a compliance risk score for each process node in the initial management process. The method also involves acquiring management parameters corresponding to the target management project, analyzing the management parameters and the compliance risk score based on a reinforcement learning model, and adjusting the process based on the analysis results to generate a project plan Gantt chart. This application can automatically generate an initial management process based on the file to be parsed and the business knowledge graph, and adjust the process based on the execution results of each process node during execution to generate a project plan Gantt chart. Combining artificial intelligence and blockchain technology, it automatically executes the project management process through smart contracts and uses reinforcement learning for process optimization, achieving automated generation and dynamic optimization of project plan Gantt charts without human intervention, thus improving the generation efficiency of project plan Gantt charts.

[0059] Please see Figure 2 , Figure 2 This is a schematic flowchart illustrating a method for generating a project plan Gantt chart, as provided in an embodiment of this application. This method can be applied to a server and, by strengthening the risk scoring and risk propagation scoring modules and combining them with preset risk coefficients for weighted fusion, can comprehensively assess the compliance risk level of a project, improving the accuracy of risk scoring and thus enhancing the compliance of project management processes.

[0060] like Figure 2 As shown, the method for generating the Gantt chart for this project specifically includes steps S201 to S204.

[0061] S201. Based on the enhanced risk scoring module, analyze the execution results to obtain project progress and compliance verification results, and analyze the project progress and the risk verification results to obtain an enhanced risk score. S202. Based on the risk propagation scoring module, analyze the business knowledge graph and the compliance verification results to obtain a risk propagation score; S203. Based on a preset risk coefficient, the enhanced risk score and the risk propagation score are weighted and fused to obtain the compliance risk score.

[0062] In one embodiment, the risk scoring model includes a reinforcement risk scoring module and a risk propagation scoring module. The risk scoring model can be trained using historical training data, wherein the reinforcement risk scoring module and the risk propagation module are trained using a joint loss function.

[0063] In one embodiment, the enhanced risk scoring module is a machine learning or data analytics-based module used to assess the risk level during project execution. A comprehensive risk score is generated by analyzing project progress and compliance verification results.

[0064] The execution results of smart contracts are stored on the blockchain. The current progress of the project is evaluated based on the execution results of smart contracts, including task completion status and adherence to time nodes.

[0065] Extract compliance verification results from the execution results. These results can include compliance tags (pass = 1, fail = 0), exceptions, etc., to determine whether the project execution process complies with the preset compliance constraints.

[0066] In one embodiment, project progress and compliance verification results are used as input state vectors for the enhanced risk scoring module. Positive rewards and negative penalties are determined based on the state vectors, and a comprehensive reward value is calculated: Comprehensive reward value = Total positive rewards - Total negative penalties. The comprehensive reward value is then mapped to an enhanced risk score within a preset range (e.g., 0-100) (lower scores indicate higher risk). Enhanced risk score = 100 - 100 × (Comprehensive reward value - Historical minimum reward) / (Historical maximum reward - Historical minimum reward).

[0067] In one embodiment, the risk propagation scoring module is used to assess the propagation of risks within a project. By analyzing dependencies in the business knowledge graph and compliance verification results, it evaluates the propagation path and scope of impact of risks, generating a risk propagation score that reflects the propagation of risks within the project, including the speed and extent of their spread.

[0068] Specifically, anomalies are extracted from the compliance verification results as initial risk sources, such as "data cross-border transmission without filing" (corresponding to the "data cross-border transmission risk" entity in the knowledge graph).

[0069] Based on the initial risk source, determine the risk type entity corresponding to the risk source in the business knowledge graph (such as the "cross-border data transmission risk" entity); the risk transmission path and probability (such as A = cross-border data transmission risk, B = customer information leakage risk, C = regulatory penalty risk, the transmission probability from A to B = 0.8, the transmission probability from B to C = 0.9); and the severity of each risk entity (such as the weight of C = 1.0, the weight of B = 0.8).

[0070] In one embodiment, the transmission paths are traversed. Specifically, the Dijkstra algorithm can be used to start from the initial risk entity, traverse all possible transmission paths, and calculate the risk value for each path by multiplying the "transmission probability × impact weight". The risk values ​​of all paths are summed to obtain the risk propagation score.

[0071] In one embodiment, a preset risk coefficient is used to adjust the weight of enhanced risk score and risk propagation score in the final compliance risk score, and can be freely set by the user according to actual needs.

[0072] Based on the preset risk coefficients, the enhanced risk score and the risk propagation score are weighted and integrated to obtain the compliance risk score.

[0073] In the above embodiments, by strengthening the risk scoring module and the risk propagation scoring module, and combining them with preset risk coefficients for weighted fusion, the compliance risk level of a project can be comprehensively assessed. This not only considers project progress and compliance but also assesses the propagation of risks, improving the accuracy of risk scoring and thus enhancing the compliance of project management processes.

[0074] Please see Figure 3 , Figure 3 This is a schematic flowchart illustrating a method for generating a project plan Gantt chart, as provided in an embodiment of this application. This method can be applied to a server to analyze project type, team capability data, and compliance risk scores using a reinforcement learning model. It generates compliance review nodes and insertion points, dynamically adjusts the project management process, and generates the project plan Gantt chart, thereby achieving dynamic optimization of the project management process and improving the compliance of the project plan Gantt chart.

[0075] like Figure 3 As shown, the method for generating the Gantt chart for this project specifically includes steps S301 to S304.

[0076] S301. Obtain the management parameters, wherein the management parameters include the project type of the target management project and the team capability data of the management team of the target management project; S302. Based on the reinforcement learning model, analyze the project type, the team capability data, and the compliance risk score to obtain analysis results, wherein the analysis results include at least one compliance review node and the insertion time point of each compliance review node; S303. Based on each compliance review node and the insertion time point, adjust the process to generate the project plan Gantt chart.

[0077] In one embodiment, the project type is the category to which the project belongs, such as software development project, construction project, marketing campaign, etc., which can be obtained through the project classification function of the project management system or extracted directly from the project plan. Different types of projects typically have different management needs and risk characteristics.

[0078] Team capability data describes the abilities and skill levels of management team members, such as their experience, professional skills, and past project success rates. This data can be obtained by acquiring team members' resumes, historical performance evaluation reports, and skill certifications, and then performing capability analysis on this information. Team capability data can help assess the team's efficiency and reliability in executing project tasks.

[0079] In one embodiment, project type, team capabilities, and compliance risk scores are integrated into the input state vector of the reinforcement learning model.

[0080] The reinforcement learning model determines the compliance review node, insertion time point, and task allocation time based on the input state vector. Specifically, the action function in the reinforcement learning model determines the compliance review node and task allocation time based on the state vector. The reward function guides the model to generate a Gantt chart that prioritizes compliance and optimizes efficiency.

[0081] For example, when the compliance risk score is ≥7, a review node will be forcibly inserted after the task (e.g., "Recordation Review" after "Cross-border Data Transfer"); fixed-cycle triggering: a regular review will be inserted every 2 weeks (e.g., on day 14, day 28); key node triggering: reviews will be inserted at milestones such as "Completion of Requirements Analysis" and "Before Test Launch". The task duration will be dynamically adjusted based on team capabilities. Task duration = industry benchmark duration × (1 - team capability coefficient), and the team capability coefficient can be extracted from team capability data.

[0082] The reward function includes compliance rewards and efficiency rewards. For example, compliance rewards could include 30 points for full coverage of all review nodes, 20 points for reduced compliance risk, 50 points for missed review nodes, and 10 points for increased risk. Efficiency rewards could include 5 points per day for completing tasks ahead of schedule and 5 points per day for delays.

[0083] In one embodiment, the process is adjusted based on the review nodes output by the reinforcement learning model, the insertion time of the review nodes, and the task allocation time of the review nodes to generate the latest project plan Gantt chart.

[0084] In the above embodiments, reinforcement learning models are used to analyze project type, team capability data, and compliance risk scores to generate compliance review nodes and insertion time points, dynamically adjust project management processes, and generate project plan Gantt charts, thereby achieving dynamic optimization of project management processes and improving the compliance of project plan Gantt charts.

[0085] Furthermore, obtaining the team capability data of the management team of the target management project includes: obtaining historical management data of the management team, analyzing the historical management data to obtain at least one capability evaluation dimension and capability scores corresponding to each capability evaluation dimension; performing hierarchical analysis on each capability evaluation dimension to obtain an importance judgment matrix; and calculating the team capability matrix based on the importance judgment matrix and the capability scores of each capability evaluation dimension, as the team capability data.

[0086] In one embodiment, historical management data refers to data accumulated by the management team in past projects, reflecting the team's management capabilities and performance. This may include project completion status, team member work efficiency, quality, and innovation capabilities, the rationality of resource allocation, resource utilization, the effectiveness of risk identification, assessment, and response, and the team's communication and coordination capabilities with internal and external stakeholders.

[0087] In one embodiment, the historical management data is preprocessed, including removing noise and irrelevant information from the data to ensure the accuracy and integrity of the data.

[0088] In one embodiment, capability evaluation dimensions are determined through data-driven approaches and expert experience. Specifically, correlation analysis is performed on preprocessed historical management data to identify indicators strongly correlated with project success rates. These indicators are then supplemented with dimensions from experts in the financial or healthcare industries to determine the final capability scoring dimensions, such as delivery capability, compliance capability, and technical capability. Strong correlation can refer to a correlation score exceeding a correlation threshold.

[0089] Extract features relevant to the competency evaluation dimensions from preprocessed historical management data. Calculate a competency score for each dimension based on the extracted features. Statistical analysis methods (such as mean, median) or machine learning methods (such as regression analysis) can be used to calculate the scores.

[0090] In one embodiment, the Analytic Hierarchy Process (AHP) is a method for multi-criteria decision analysis. By constructing a hierarchical model, complex problems are decomposed into multiple levels and factors, and then the relative importance of each factor is determined through pairwise comparisons.

[0091] Specifically, a hierarchical structure is constructed, including: the goal layer (comprehensive assessment of team capabilities); the criteria layer (capability evaluation dimensions); and the solution layer (specific scores for each scoring dimension).

[0092] Construct an importance judgment matrix, obtain expert scores or team discussions, and compare each capability evaluation dimension of the criteria layer pairwise to determine their relative importance. The comparison results are represented by a matrix called the importance judgment matrix. For example, assume... The importance matrix for delivery capability, compliance capability, and technical capability is as follows:

[0093] In the matrix, 1 in the first row indicates that the abilities are equally important, and 3 indicates that... Ability ratio Ability is slightly more important, 5 indicates Ability ratio Ability is much more important.

[0094] A consistency check is performed on the importance judgment matrix to ensure its consistency. The consistency index CI = (largest eigenvalue of the importance judgment matrix - number of dimensions) / (number of dimensions - 1). If the consistency ratio (CR = CI / RI, where RI is the average random consistency index) is less than a certain threshold, the matrix is ​​considered to have satisfactory consistency.

[0095] In one embodiment, the geometric mean method is used to calculate the weights of each dimension. Specifically, the geometric mean of the elements in each row of the importance judgment matrix is ​​calculated, and the geometric mean is normalized to obtain the weights of each capability evaluation dimension.

[0096] The ability scores for each capability evaluation dimension are multiplied by their corresponding weights to obtain weighted scores for each capability evaluation dimension. These weighted scores are then summed to obtain the overall team capability score. The capability evaluation dimensions, capability scores, importance weights, weighted scores, and overall capability score are represented as a matrix as the team capability data.

[0097] In the above implementation, by acquiring historical management data of the management team, analyzing the data to obtain capability evaluation dimensions and capability scores, and using the analytic hierarchy process (AHP) to construct an importance judgment matrix, the team capability matrix can be calculated, enabling a comprehensive assessment of the management team's capabilities. This approach not only considers multiple capability evaluation dimensions but also uses the AHP to determine the relative importance of each dimension, providing scientific decision support for project management and improving the rationality of the project management process.

[0098] Please see Figure 4 , Figure 4 This embodiment of the present application provides a schematic block diagram of a project plan Gantt chart generation apparatus, which is used to execute the aforementioned project plan Gantt chart generation method. The project plan Gantt chart generation apparatus can be configured on a server.

[0099] like Figure 4 As shown, the project plans to use a Gantt chart generation device 400, which includes: The initial management process generation module 401 is used to obtain the file to be parsed and the business knowledge graph corresponding to the target management project, and process the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project. The compliance risk score acquisition module 402 is used to generate and execute a smart contract based on the initial management process, obtain the execution results of each process node in the initial management process, and perform risk scoring on the execution results based on the risk scoring model to obtain the compliance risk score of each process node in the initial management process. The project plan Gantt chart generation module 403 is used to obtain the management parameters corresponding to the target management project, analyze the management parameters and the compliance risk score based on the reinforcement learning model, and adjust the process according to the analysis results to generate the project plan Gantt chart.

[0100] Furthermore, the risk scoring model includes a reinforced risk scoring module and a risk propagation scoring module, and the compliance risk score acquisition module 402 includes: The enhanced risk scoring unit is used to analyze the execution results based on the enhanced risk scoring module to obtain project progress and compliance verification results, and to analyze the project progress and the risk verification results to obtain an enhanced risk score. The risk propagation score acquisition unit is used to analyze the business knowledge graph and the compliance verification results based on the risk propagation score module to obtain a risk propagation score. The compliance risk score acquisition unit is used to perform weighted fusion of the enhanced risk score and the risk propagation score based on a preset risk coefficient to obtain the compliance risk score.

[0101] Furthermore, the project plan Gantt chart generation module 403 includes: A management parameter acquisition unit is used to acquire the management parameters, wherein the management parameters include the project type of the target management project and the team capability data of the management team of the target management project; The analysis result acquisition unit is used to analyze the project type, the team capability data and the compliance risk score based on the reinforcement learning model to obtain analysis results, wherein the analysis results include at least one compliance review node and the insertion time point of each compliance review node; The Gantt chart generation unit is used to generate the project plan Gantt chart by adjusting the process based on each compliance review node and the insertion time point.

[0102] Furthermore, the management parameter acquisition unit includes: The historical management data analysis subunit is used to acquire the historical management data of the management team and analyze the historical management data to obtain at least one capability evaluation dimension and capability scores corresponding to each capability evaluation dimension. The importance judgment matrix is ​​obtained by sub-units, which are used to perform hierarchical analysis on each of the aforementioned capability evaluation dimensions to obtain the importance judgment matrix; The team capability data acquisition subunit is used to calculate the team capability matrix based on the importance judgment matrix and the capability scores of each capability evaluation dimension, as the team capability data.

[0103] Furthermore, the initial management process generation module 401 includes: The file parsing unit is used to parse the file to be parsed based on a preset large language model to obtain the first compliance constraint and the first process node. The information extraction unit is used to extract information from the business knowledge graph to obtain the second compliance constraint and the second process node. The data integration unit is used to integrate the first compliance constraint and the second compliance constraint to generate compliance constraint, and to analyze and integrate the first process node and the second process node to generate process nodes to be processed. The condition mapping unit is used to map the compliance constraints to the process nodes to be processed, thereby obtaining the initial management process.

[0104] Further, the condition mapping unit includes: The data classification subunit is used to classify the compliance constraints according to data modalities to obtain text data, structured data, and knowledge graph data; The data mapping subunit is used to map the text data, the structured data, and the knowledge graph data to a shared vector space to obtain a data vector with a unified standard. The condition mapping subunit is used to perform sparse attention processing on the data vector and the process node to be processed based on the cross attention layer in the preset self-attention model, so as to match and map the compliance constraints with the process node to be processed to obtain the initial management process.

[0105] Furthermore, the initial management process generation module 401 includes: The business document parsing unit is used to parse the business documents corresponding to the target management project based on the local large language model of each project management party, and obtain at least one key entity that matches the target management project. The logical relationship acquisition unit is used to identify the relationships between the key entities based on the dependency parsing model and obtain the logical relationships between the key entities. The business knowledge graph generation unit is used to aggregate the received key entities and the logical relationships between them based on a federated learning algorithm to generate the business knowledge graph.

[0106] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the above-described apparatus and modules can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0107] The aforementioned device can be implemented as a computer program, which can be used in, for example... Figure 5 It runs on the computer device shown.

[0108] Please see Figure 5 , Figure 5 This is a schematic block diagram illustrating the structure of a computer device according to an embodiment of this application. The computer device may be a server.

[0109] See Figure 5 The computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0110] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any method for generating a project plan Gantt chart.

[0111] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0112] Internal memory provides an environment for the execution of computer programs on non-volatile storage media. When executed by a processor, the computer program enables the processor to perform any method for generating a project plan Gantt chart.

[0113] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0114] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0115] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Obtain the file to be parsed and the business knowledge graph corresponding to the target management project, and process the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project; Based on the initial management process, a smart contract is generated and executed to obtain the execution results of each process node in the initial management process. The execution results are then risk-scored based on a risk scoring model to obtain the compliance risk score of each process node in the initial management process. Obtain the management parameters corresponding to the target management project, analyze the management parameters and the compliance risk score based on the reinforcement learning model, adjust the process according to the analysis results, and generate a project plan Gantt chart.

[0116] In one embodiment, the risk scoring model includes a reinforced risk scoring module and a risk propagation scoring module. When the processor performs risk scoring on the execution results based on the risk scoring model to obtain the compliance risk scores for each process node in the initial management process, it is used to: Based on the enhanced risk scoring module, the execution results are analyzed to obtain project progress and compliance verification results, and the enhanced risk score is obtained by analyzing the project progress and the risk verification results. Based on the risk propagation scoring module, the business knowledge graph and the compliance verification results are analyzed to obtain a risk propagation score; Based on a preset risk coefficient, the enhanced risk score and the risk propagation score are weighted and fused to obtain the compliance risk score.

[0117] In one embodiment, when the processor acquires the management parameters corresponding to the target management project, analyzes the management parameters and the compliance risk score based on a reinforcement learning model, adjusts the process according to the analysis results, and generates a project plan Gantt chart, it is used to: Obtain the management parameters, wherein the management parameters include the project type of the target management project and the team capability data of the management team of the target management project; The reinforcement learning model is used to analyze the project type, the team capability data, and the compliance risk score to obtain analysis results, wherein the analysis results include at least one compliance review node and the insertion time point of each compliance review node; Based on the compliance review nodes and the insertion time points, the process is adjusted to generate the project plan Gantt chart.

[0118] In one embodiment, when the processor acquires the team capability data of the management team of the target management project, it is configured to: Obtain historical management data of the management team and analyze the historical management data to obtain at least one capability evaluation dimension and capability scores corresponding to each capability evaluation dimension; A hierarchical analysis was performed on each of the aforementioned capability evaluation dimensions to obtain an importance judgment matrix; Based on the importance judgment matrix and the capability scores of each capability evaluation dimension, the team capability matrix is ​​calculated and used as the team capability data.

[0119] In one embodiment, when the processor processes the file to be parsed and the business knowledge graph to obtain the initial management process corresponding to the target management project, it is configured to: The file to be parsed is parsed based on a preset large language model to obtain the first compliance constraint and the first process node; Information is extracted from the business knowledge graph to obtain the second compliance constraint and the second process node; The first compliance constraint and the second compliance constraint are integrated to generate compliance constraint conditions. The first process node and the second process node are analyzed and integrated to generate process nodes to be processed. The compliance constraints are mapped to the process nodes to be processed to obtain the initial management process.

[0120] In one embodiment, when the processor maps the compliance constraints to the process node to be processed to obtain the initial management process, it is configured to: The compliance constraints are classified according to data modalities to obtain text data, structured data, and knowledge graph data; The text data, the structured data, and the knowledge graph data are mapped to a shared vector space to obtain a data vector with a unified standard. Based on the cross-attention layer in the preset self-attention model, sparse attention processing is performed on the data vector and the process node to be processed to match and map the compliance constraints with the process node to be processed, thereby obtaining the initial management process.

[0121] In one embodiment, when the processor acquires the business knowledge graph corresponding to the target management project, it is used to: Based on the local large language model of each project management party, the business files corresponding to the target management project are parsed to obtain at least one key entity that matches the target management project. Based on the dependency parsing model, the relationship between each key entity is identified to obtain the logical relationship between each key entity. The business knowledge graph is generated by aggregating the received key entities and the logical relationships between them using a federated learning algorithm.

[0122] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement any of the project plan Gantt chart generation methods provided in the embodiments of this application.

[0123] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0124] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method of generating a Gantt chart of a project plan, characterized by, The method comprises the following steps: obtaining a to-be-analyzed file corresponding to a target management project and a business knowledge graph, and processing the to-be-analyzed file and the business knowledge graph to obtain an initial management process corresponding to the target management project; generating an intelligent contract based on the initial management process and executing the intelligent contract to obtain an execution result of each process node in the initial management process, and performing risk scoring on the execution result based on a risk scoring model to obtain a compliance risk score of each process node in the initial management process; obtaining management parameters corresponding to the target management project, analyzing the management parameters and the compliance risk score based on a reinforcement learning model, and adjusting the process according to the analysis result to generate a project plan Gantt chart.

2. The method of claim 1, wherein, The risk scoring model comprises a reinforcement risk scoring module and a risk propagation scoring module, and the risk scoring based on the risk scoring model and the execution result to obtain the compliance risk score of each process node in the initial management process comprises: analyzing the execution result based on the reinforcement risk scoring module to obtain project progress and compliance verification results, and analyzing the project progress and the risk verification results to obtain a reinforcement risk score; analyzing the business knowledge graph and the compliance verification results based on the risk propagation scoring module to obtain a risk propagation score; based on a preset risk coefficient, the reinforcement risk score and the risk propagation score are weighted and fused to obtain the compliance risk score.

3. The method of claim 1, wherein, The method comprises the following steps: obtaining the management parameters, wherein the management parameters comprise a project type of the target management project and team capability data of a management team of the target management project; analyzing the project type, the team capability data and the compliance risk score based on the reinforcement learning model to obtain an analysis result, wherein the analysis result comprises at least one compliance review node and an insertion time point of each compliance review node; based on each compliance review node and the insertion time point, the process is adjusted to generate the project plan Gantt chart.

4. The method of claim 3, wherein, The method comprises the following steps: obtaining historical management data of the management team, and analyzing the historical management data to obtain at least one capability evaluation dimension and a capability score corresponding to each capability evaluation dimension; analyzing each capability evaluation dimension to obtain an importance judgment matrix; based on the importance judgment matrix and the capability score of each capability evaluation dimension, the team capability matrix is calculated and obtained as the team capability data.

5. The method of claim 1, wherein, The method comprises the following steps: based on a preset large language model, the to-be-analyzed file is analyzed to obtain a first compliance constraint condition and a first process node; information extraction is performed on the business knowledge graph to obtain second compliance constraint conditions and second process nodes; the first compliance constraint conditions and the second compliance constraint conditions are integrated to generate compliance constraint conditions, and the first process nodes and the second process nodes are analyzed and integrated to generate to-be-processed process nodes; the compliance constraint conditions are mapped to the to-be-processed process nodes to obtain the initial management process.

6. The method of claim 5, wherein, The compliance constraint conditions are mapped to the to-be-processed process nodes to obtain the initial management process, including: The compliance constraint conditions are classified according to data modalities to obtain text data, structured data, and knowledge graph data; the text data, the structured data, and the knowledge graph data are mapped to a shared vector space to obtain a unified standard data vector; the data vector and the to-be-processed process nodes are subjected to sparse attention processing based on a cross-attention layer in a preset self-attention model, so as to match and map the compliance constraint conditions to the to-be-processed process nodes, and obtain the initial management process.

7. The method of generating a Gantt chart of a project plan according to any one of claims 1 to 6, characterized in that, The business knowledge graph corresponding to the target management project is obtained, including: The business files corresponding to the target management project are parsed based on a local large language model of each project management party to obtain at least one key entity matched with the target management project; The logical relationships between the key entities are obtained based on a dependency syntax analysis model. The business knowledge graph is generated by aggregating the key entities and the logical relationships between the key entities based on a federated learning algorithm.

8. An apparatus for generating a Gantt chart of a project plan, characterized by including: An initial management process generation module is configured to obtain to-be-processed files and a business knowledge graph corresponding to a target management project, and process the to-be-processed files and the business knowledge graph to obtain an initial management process corresponding to the target management project; A compliance risk score obtaining module is configured to generate an intelligent contract based on the initial management process and perform the intelligent contract to obtain execution results of each process node in the initial management process, and perform risk scoring on the execution results based on a risk scoring model to obtain compliance risk scores of each process node in the initial management process; A project plan Gantt chart generation module is configured to obtain management parameters corresponding to the target management project, analyze the management parameters and the compliance risk scores based on a reinforcement learning model, and generate a project plan Gantt chart based on the analysis results.

9. A computer device, comprising: The computer device includes a memory and a processor; The memory is configured to store a computer program; The processor is configured to execute the computer program and implement the project plan Gantt chart generation method of any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program causes the processor to implement the project plan Gantt chart generation method of any one of claims 1 to 7 when executed by the processor.

Images